I think you're starting with a model that's too complex.
I usually start my studies in a very simple way. The idea is to have a model that "works" and then each time you add a hypothesis that allows you to get closer to reality and you check if "it works" and if it is coherent. Once all your hypotheses have been made, then and only then do you densify your mesh for the final calculation.
Note that the connectors add many additional equations and therefore a significant calculation time.
If I were you, I would start by putting a "global solidarity contact" and I would make a first calculation. then you refine as best you can...
At least that's my way of proceeding, because otherwise you don't know where you have it anymore and it's a hassle to know what works or not.
When you say try to have only a global solidarity contact, what does it mean because I have my platform which is in welded mechanical element so he tells me his :
Ok well I did as you told me and visibly for the moment everything works.
I'll see the results because for the moment in the first simulation it tells me that there is no deformation in the carriage (personally I only half believe it) and I'll add elements and we'll see if it brings a change if I ever find a solution I'll come back and have a look around here:)